Early Cognitive Computer Vision

Abstract This chapter outlines computational models for the first stages of visual cognition. For both biological and technical systems, we are examining which architectural components are necessary in such systems, and how experience can be acquired and used to steer perceptual interpretation. Since human perception has evolved to interpret the structure of the world around us, a necessary boundary condition of the vision system must be the common statistics of natural images. Searching for generality, it is observed that a limited set of physical laws of image formation will impress common statistics on the images offered to the eye as sensory input. The physical laws are largely scene and domain independent, as they cover the universally applicable laws of light reflectance from materials. The chapter focuses on the physical and statistical constrains in the sensory input, and how this can be exploited to construct cognitive vision systems. Visual cognition may be based on a weak description of the important features in the scene, as long as mutual correspondence between observation and objects in the world is maintained. For such a computational theory, the first few steps will be outlined: visual measurement, invariant representation, and focal attention.